Collaborative Decision Support Systems for Facility Management
Total Page:16
File Type:pdf, Size:1020Kb
FE Collaborative Decision Support Systems for Facility Management Arthur J. Chapman President COM Technologies. Inc 2975 McMillan Ave. Suite 272 San Luis Obispo, CA 93401 Email: [email protected] Abstract Agent-based collaborative decision support is a methodology of utilizing domain specific intelligent systems, interacting in a common environment, to partner with one or more human decision makers to reach a consensus solution to a complex problem. An example is the recently developed Collaborative Infrastructure Assessment Tool (CIAT) that provides a collaborative planning facility management tool in support of military pier and port management. This methodology is applicable to many similar dynamic facility management problems where the complexity of issues and the number of decision makers result in the need for domain specific agents, a common view of the data, and the need to reach a consensus solution. Keywords agents, collaborative planning, consensus solution, decision support, facilities management, infrastructure assessment, knowledge base, problem solving, Decision Support Data Decision support is one of the most talked about and least implemented topics in the computer world today. It is generally built around a data warehousing component where corporate or institutional data are centralized and sanitized for general distribution. This is coupled with a variety of presentation programs that allow the slicing and dicing of data fields with the goal of providing users with every imaginable presentation option. The data collection and presentation process has unfortunately absorbed most corporate or institutional organizations and makes difficult their ability to evolve into decision support, whereby data are combined with operational rules and structured into a collaborative decision making system. A recent survey of 700 information systems (IS) executives from a worldwide cross-section of large corporations found their foremost concern was not decision support but data and data movement, in particular, data hoarding, data 71 compatibility among operating systems, and network performance for data movement. Particular concerns were "...critical information dispersed outside the data center, where it can't be easily accessed and managed, weakens customer service, lowers productivity and hurts a company's ability to compete.....most IS executives run into roadblocks when trying to share data because of the multitude of computer operating systems in use in their companies, and the impact that moving data over networks has on the performance of their main production systems" (The HP Chronicle, December 1997). Their focus is on data productivity and the perception that there is a direct relationship between the availability of data and the company "bottom line" (Le., profitability). For those reasons, industrial trends are for computer services to become more compatible and faster, and therefore will soon make feasible the direct dependence, in 'real-time', on the information buried in the huge piles of data these IS executives are worried about. Unfortunately, their data concerns overshadow a much larger issue. As data become more readily available for operational and strategic purposes, users are demanding systems that allow help in understanding, interpreting, digesting, and incorporating the data into their decision making roles. In being consumed with the process of data collecting, IS executives seem not to be addressing appropriate data use and the decision making process. For example, in the area of facility management, what would it mean to have immediate access to 10 years of maintenance records for all air-conditioning equipment for an institution's 125 buildings. Maybe nothing. What would it mean to have fully detailed as-built drawings of every building for a large corporation and have them on a super computer-aided drawing system that is 'web' compatible and accessible to all employees both on an off their site. Maybe nothing. Unfortunately, many organizations are moving forward with weakly founded data collection plans hoping, as in the movie Field of Dreams, "build (pile) it and they will come (use it)". Decision Support Data Needs Data should be viewed as one of two primary support components for decision making. The other being decision support rules. Decision making can occur when data are combined with rules. The result may then be called a knowledge base. Data that have no basis for use by rules should be archived and the expense of such data collection questioned. For example, one might have the rule that an air-conditioning system will be serviced after 8,000 hours of operation. That rule could then be part of a computer-based maintenance agent that tracks air-conditioning operational data. When service hours exceed a specified limit, the rule 'fires' and alerts the appropriate personnel. The maintenance agent could also anticipate when multiple buildings are to have their air-conditioning systems serviced and alert management of an impending 72 • workload crisis. Such service data have a reason for being and clearly would justify their existence. A service agent is a rule-based computer program dedicated to monitoring and advising in a narrow domain. A 'real-time' example might be a weather agent monitoring weather data and advising, either a human decision maker or another agent, on appropriate precautions for bUilding operations. A building use agent might monitor building activity levels and prepare for increases in maintenance needs. A maintenance agent might collaborate with the building use agent and propose a schedule of maintenance activities. A project schedUling agent might consider and advise on the time overlapping of trades and facility use needs. An agent is the domain specific 'engine' that combines rules and data. Groups of agents, including the human agents, form a Decision Making Team (DMT). Agent-Based Collaborative Decision Making (ICOM) Advances in agent-based software systems incorporating rule-based expert systems that assist human decision makers by monitoring user actions, evaluating consequences, proposing solution strategies, and inferring alternative solutions, represent a promising direction for assisting in the solution to complex facility management problems. In such distributed, cooperative systems, mUltiple intelligent agents collaborate among themselves and with the DMT to improve the quality of the planning effort. The Integrated Cooperative Decision Making (I COM) architecture developed jointly by the CAD Research Center (CADRC) at California Polytechnic State University and COM Technologies, Inc., provides a distributed computer-based framework for cooperative decision support systems. It comprises a data blackboard that is capable of interacting with multiple distributed agents, and addresses the support needs of a complex planning situation in terms of: '* A high level internal representation of real world objects that are commonly used by human decision makers in a particular application area. * Databases containing factual technical information, reference data, and prototype models of knowledge that pertains to the problem under consideration. * Expert agents capable of validating decisions and predicting the impact of solutions in one area on solutions in related areas. The distributed agent computing model is capable of emulating and extending basic human problem solving strategies for the solution of complex planning problems. Human problem solving generally involves collaboration among DMT experts, such as building engineers, maintenance personnel, construction managers, and so on, depending on the particular problem. By encapsulating 73 - knowledge from human domain experts into computer-based agents and allowing these agents to interact with each other and with human users, and by allowing this interaction to occur in parallel through the use of computer networks, such systems can significantly increase the productivity of decision making. The CADRe and COM Technologies, .Inc. have implemented this ICOM methodology within a number of fielded projects (Pohl et ai, 1997). A CIAT Implementation of the ICOM Model The Collaborative Infrastructure Assessment Tool (CIAT) is a collaborative planning facility management tool that provides support to a group of decision makers each with unique responsibilities but working for a common goal. The application to be discussed is a DMT in support of US Navy pier and port management. Many US Navy ports are faced with the common problem of coordinating the movement and positioning of large warfare and support ships into selected berth locations. The issues of this problem go far beyond the simple matching of ships to avai.lable berths. Each naval pier facility has a unique set of characteristics including, utility availability, berth drafts, loading capability, width, scheduled maintenance activities, and so on. In addition, each pier has a unique relationship with port facilities based on location. For example. some may have more desirable access to parking, restaurants, recreational facilities, and so on. In addition, as ships prepare for arrival they transmit requests for services that may involve unexpected facility uses. These issues result in a complex problem solVing situation, requiring a DMT, and a situation that is well suited to the collaborative planning environment as provided by the ICOM architecture. CIAT is currently being used as the primary tool to place ships at the US Naval Station in San